HA test failed

[mixmouse]

We have two RS/6000 servers running as HA cluster. Both of them have DB2 and MQSeries application installed.
We did HA take over test a few days ago, but the test failed because the service IP address could not start up.

Following are our procedures:
(Before test, cluster services on two servers were running)
1. From server A: setup cluster services mode to takeover and stop cluster service.
Result: 1) cluster services on server B started up.
2) Ping server B's standby IP - OK.
3) Ping HA service IP - failed.
4) DB2 HA application server started OK.
5) MQSeries HA application server started OK.

2. Stop and started cluster services on server B mannually.
Result: 1) cluster services on server B started up.
2) Ping server B's standby IP - OK.
3) Ping HA service IP - OK.
4) DB2 HA application server started OK.
5) MQSeries HA application server started OK.

We checked all HA logs but could not find any error message on both servers.

Is there anyone can give me some suggestion?

Thanks

 

[jprabaker]

What is the content of the cm.log and cluster.log files?

 

[brigode]

Did "A" give up it's service address?

Which order Does failover take? File System / Network Adapters OR Reverse?

I would look very close at event log and the cmlog. Did it even try?

Have you been succesful before?

What does clstat look like?

Sounds to me like either a sset up issue or there is a hanging event from a prior failure.

 

[gheist]

these tests succeed only if you still run rlogind via inetd (i usually stop it when clustering system comes up and start it when starting cluster service)